yolov8官方c++代码vs2019onnxruntime推理失败问题

v8定位模型，使用官方代码，推理结果为杂乱无章的定位框或者框都在左上角。

调试后发现是后处理代码处，对输出头的形状处理不正确，已修正

位置原代码inference.cpp第224行，后处理部分

原代码：

        int strideNum = outputNodeDims[1];//8400
        int signalResultNum = outputNodeDims[2];//84
        std::vector<int> class_ids;
        std::vector<float> confidences;
        std::vector<cv::Rect> boxes;
        cv::Mat rawData;
        if (modelType == YOLO_DETECT_V8)
        {
            // FP32
            rawData = cv::Mat(strideNum, signalResultNum, CV_32F, output);
        }
        else
        {
            // FP16
            rawData = cv::Mat(strideNum, signalResultNum, CV_16F, output);
            rawData.convertTo(rawData, CV_32F);
        }


        float* data = (float*)rawData.data;

修改后代码：

        int strideNum = outputNodeDims[2];//8400
        int signalResultNum = outputNodeDims[1];//84
        std::vector<int> class_ids;
        std::vector<float> confidences;
        std::vector<cv::Rect> boxes;
        cv::Mat rawData;
        cv::Mat transposed_output;
        if (modelType == YOLO_DETECT_V8)
        {
            // FP32
            //rawData = cv::Mat(strideNum, signalResultNum, CV_32F, output);
            rawData = cv::Mat(signalResultNum, strideNum, CV_32F, output);
            cv::transpose(rawData, transposed_output);
        }
        else
        {
            // FP16
            rawData = cv::Mat(signalResultNum, strideNum, CV_16F, output);
            rawData.convertTo(rawData, CV_32F);
            cv::transpose(rawData, transposed_output);
        }


        //float* data = (float*)rawData.data;
        float* data = (float*)transposed_output.data;

修改后定位框准确无误